CtrlK
BlogDocsLog inGet started
Tessl Logo

task-orchestration-governor

Govern the JM Labs persistent task filesystem and intake workflow. Use whenever the task mentions `workspaces/tasks`, `tasklog.md`, `rag-memory-*`, task intake, task close, task resume, ambiguous inputs, session-state orchestration, or the rule that every new work input must be recoverable from disk.

SKILL.md
Quality
Evals
Security

Task Orchestration Governor

Use this skill to keep the task filesystem deterministic, recoverable, and strictly local.

First reads

  1. Read /Users/deonto/jm-claude-desktop-workspace-core/CLAUDE.md.
  2. Read /Users/deonto/jm-claude-desktop-workspace-core/session-state.template.json.
  3. If the task touches portability, read /Users/deonto/jm-claude-desktop-workspace-core/contracts/shared-sync-allowlist.json.

Governing rules

  1. Route all task-local memory to workspaces/tasks/.
  2. Keep tasklog.md generated and open-only.
  3. Resolve every new input as new, resume, or ambiguous.
  4. On ambiguity, preserve the input and stop for confirmation instead of mixing contexts.
  5. Never close a task without explicit confirmation and completed DoD.

Required checks

  • Verify the contract files exist: task.md, status.json, definition-of-done.md, conversation-log.md, rag-memory-*.md.
  • Verify session-state.json carries current task and ambiguity fields.
  • Verify workspaces/** and tasklog.md remain local-only and ignored.

Output format

Resolution path: <new | resume | ambiguous | repair>
Files or scripts: <what must change or run>
Boundary check: <why this stays local or portable>
Verification: <commands or evidence>

Hard boundaries

  • Do not store task memory outside workspaces/tasks/.
  • Do not hand-edit tasklog.md as a durable narrative log.
  • Do not claim resolution quality if the intake path did not write a rag-memory-*.
Repository
JaviMontano/jm-claude-desktop-workspace-core-backup
Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.